From 69c9d9207a658bd048fbf54016a5fa4b824a1531 Mon Sep 17 00:00:00 2001
From: Arne Georg Gleditsch <argggh@taniquetil.(none)>
Date: Wed, 12 Aug 2009 22:39:46 +0200
Subject: Make sure purposely skipped files are not marked as pending.

---
 lxr-genxref | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/lxr-genxref b/lxr-genxref
index 782ad9f..599f202 100755
--- a/lxr-genxref
+++ b/lxr-genxref
@@ -280,16 +280,22 @@ sub reference_file($$$) {
 sub hash_file($$$) {
     my ($file, $fileid, $rels) = @_;
 
-    return 0 if
-	defined($context->config->{'search_size_limit'}) and
+    if (defined($context->config->{'search_size_limit'}) and
 	$context->config->{'search_size_limit'} > 0 and
-	$file->size > $context->config->{'search_size_limit'};
+	$file->size > $context->config->{'search_size_limit'})
+    {
+	$index->to_hash($fileid);
+	$index->to_reference($fileid);
+	return 0;
+    }
 
     my $docid;
     if ($index->to_hash($fileid)) {
 	my $handle;
 	sysopen($handle, $file->phys_path, 0) || die($!);
 	unless (-T $handle) {
+	    # Non-text file.  Mark as referenced as well.
+	    $index->to_reference($fileid);
 	    $handle->close();
 	    return 0;
 	}
@@ -323,7 +329,10 @@ sub hash_file($$$) {
     }
     else {
 	$docid = $index->get_hashed_document($fileid);
-	return 0 unless $docid;
+	unless ($docid) {
+	    $index->to_reference($fileid);
+	    return 0;
+	}
 	my $doc = $hash->get_document($docid);
 	if (reference_file($file, $fileid, $doc)) {
 	    $hash->save_document($docid, $doc);
-- 
cgit v1.2.3


From 65c53ad4bf0589c890f9c07383858f244c02c35a Mon Sep 17 00:00:00 2001
From: Arne Georg Gleditsch <argggh@taniquetil.(none)>
Date: Wed, 12 Aug 2009 22:40:17 +0200
Subject: Don't do textual manipulations on references...

---
 lib/LXRng/Index/DBI.pm | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/LXRng/Index/DBI.pm b/lib/LXRng/Index/DBI.pm
index 8fcb391..eeb6a83 100644
--- a/lib/LXRng/Index/DBI.pm
+++ b/lib/LXRng/Index/DBI.pm
@@ -471,7 +471,8 @@ sub get_symbol_usage {
 
     my %rlines;
     foreach my $r (@$res) {
-	$rlines{$$r[0]} = [$$r[1] =~ /(\d+),?/g];
+	$rlines{$$r[0]} = ref($$r[1]) eq 'ARRAY' 
+	    ? $$r[1] : [$$r[1] =~ /(\d+),?/g];
     }
 
     return \%rlines;
-- 
cgit v1.2.3


From 40c6f593f54c8023d6ed74e695de6b3a56a74bad Mon Sep 17 00:00:00 2001
From: Arne Georg Gleditsch <argggh@taniquetil.(none)>
Date: Wed, 12 Aug 2009 22:40:51 +0200
Subject: Increase flush intervals somewhat.

---
 lib/LXRng/Index/PgBatch.pm | 2 +-
 lib/LXRng/Search/Xapian.pm | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/LXRng/Index/PgBatch.pm b/lib/LXRng/Index/PgBatch.pm
index c1b30eb..7bd67e6 100644
--- a/lib/LXRng/Index/PgBatch.pm
+++ b/lib/LXRng/Index/PgBatch.pm
@@ -52,7 +52,7 @@ sub transaction {
 	# Only occasional synchronization if we're inside another
 	# transaction.
 	# TODO: Check fill grade of caches and flush based on that.
-	if ($self->{'writes'}++ % 491 == 0) {
+	if ($self->{'writes'}++ % 3259 == 0) {
 	    $self->flush();
 	    $self->dbh->commit();
 	}
diff --git a/lib/LXRng/Search/Xapian.pm b/lib/LXRng/Search/Xapian.pm
index 014d57a..5230728 100644
--- a/lib/LXRng/Search/Xapian.pm
+++ b/lib/LXRng/Search/Xapian.pm
@@ -81,7 +81,7 @@ sub add_document {
     }
     my $doc_id = $self->wrdb->add_document($doc);
     $self->{'writes'}++;
-    $self->flush() if $self->{'writes'} % 499 == 0;
+    $self->flush() if $self->{'writes'} % 3271 == 0;
     return $doc_id;
 }
 
-- 
cgit v1.2.3